Memento

Memento Flashcards: intelligent flashcard learning powered by AI and FSRS, with your data stored locally for complete privacy.

Comment

Inspiration

I’ve always loved the efficiency of spaced-repetition tools, but the options available felt either too bloated, too limited, or too dependent on cloud services. I wanted a flashcard app that was lightweight, private, offline-first, and flexible, while still offering powerful study tools and modern AI-powered features.

Memento Flashcards was born from that idea: a simple, organized, personal learning system designed to help anyone retain knowledge effectively without sacrificing privacy or usability.

What it does

Memento Flashcards is a desktop app that helps users learn more efficiently through:

  • Organized study structures with folders, subfolders, decks, and cards
  • Three study modes, including an FSRS (Free Spaced Repetition Scheduler) based review system
  • AI-powered flashcard generation using the OpenRouter API
  • Full control over creating, renaming, deleting, and managing study content
  • 100% offline storage, keeping all data local in an SQLite database

The app launches in a browser window and provides a clean, intuitive interface for both studying and managing learning material.

How I built it

Memento was built using a modern, lightweight stack:

  • Frontend: React
  • Backend: FastAPI
  • Database: SQLite for fast local storage
  • Algorithm: FSRS implementation for intelligent spaced-repetition scheduling
  • AI Integration: OpenRouter API to generate flashcards automatically from user prompts

The app is packaged as a Windows executable. On first launch, it automatically creates a local SQLite database at: C:\Users\AppData\Local\SrijanRavisankar\MementoFlashcards\memento_flashcard_db.sqlite3

Challenges I ran into

Building Memento came with several hurdles:

  • Integrating the FSRS algorithm required careful state tracking and math to get scheduling right.
  • Managing nested structures (folders → subfolders → decks → cards) while keeping the UI simple took extensive iteration.
  • Packaging FastAPI + React into a clean Windows executable was trickier than expected.
  • AI flashcard generation required prompt tuning and rate handling to ensure consistent results.
  • Browser-based launching occasionally triggered Windows security warnings, which required clear installation guidance.

Accomplishments that I am proud of

  • Implementing a fully offline spaced-repetition system with FSRS state transitions
  • Building a smooth, nested organizational system for folders, decks, and cards
  • Creating a clean, easy-to-use UI that works directly in the browser
  • Integrating AI-powered flashcard generation seamlessly
  • Delivering a single-download, local-only tool that prioritizes user privacy

What I learned

Throughout the project, I learned:

  • How to implement and tune the FSRS scheduling algorithm
  • Best practices for React–FastAPI communication
  • How to structure offline-first apps with local persistence
  • The importance of thoughtful UX when dealing with complex data hierarchies
  • Practical experience working with AI APIs, prompt design, and error-handling
  • How to package and distribute desktop applications cleanly

What's next for Memento

The future roadmap includes:

  • Cross-platform support (Linux, macOS, and possibly mobile)
  • Custom themes, including dark mode
  • Search and filtering across folders and decks
  • Advanced analytics on study progress

Hackathon New Progress (10–16 Nov 2025):

The following progress was made between 10-16 November 2025:

  • Feature 1: Added comprehensive documentation: Created detailed setup guides, usage instructions, and made feature descriptions clearer to make the app accessible for new users.
  • Feature 2: Recorded and published a full video tutorial: Produced an end-to-end walkthrough explaining how to install the app, create decks, study using FSRS, and generate AI flashcards.

Built With

Share this project:

Updates